Photonic Integrated Circuits (PICs) promise to revolutionize computing, routing, and quantum information processing by moving data with light instead of electricity. However, managing light requires reconfigurability—the ability to dynamically route, modulate, or delay optical signals on a chip.
Traditionally, this tuning relies on volatile mechanisms like the thermo-optic effect, which requires a continuous injection of electrical power to maintain a specific state. To achieve true energy efficiency, researchers have turned to Phase-Change Materials (PCMs). These materials can switch between amorphous and crystalline atomic structures, offering a "set-and-forget" non-volatile state retention.
Yet, traditional PCMs come with a severe catch: material-induced optical loss.
The Physics of Parasitic Absorption
To understand why traditional PCMs struggle in photonics, we have to look at their electronic band structures. The most widely studied PCM, Germanium-Antimony-Tellurium ($\text{Ge}_2\text{Sb}_2\text{Te}_5$ or GST), was originally optimized for rewritable optical discs and electronic phase-change memory. In these applications, a massive contrast in electrical conductivity or optical reflectivity between states is highly desirable.
However, as Rui Chen et al. note in Nonvolatile electrically controllable broadband switches based on phase-change material GST, this contrast stems from a narrow electronic bandgap. When GST transitions into its crystalline state, its bandgap narrows further, causing it to behave almost like a metal.
Traditional PCM (e.g., GST)
[ Photon Energy (1550nm) ] ---> [ Narrow Bandgap ] ---> High Absorption (Loss)
Wide-Bandgap PCM (e.g., Sb2S3)
[ Photon Energy (1550nm) ] ---> [ Wide Bandgap ] ---> Zero Absorption (Transparent)
When telecommunication light (typically at a wavelength of 1550 nm, corresponding to a photon energy of ~0.8 eV) passes through crystalline GST, the photons possess more than enough energy to excite electrons across this narrow gap. This triggers intense band-to-band absorption. The imaginary part of the refractive index—the extinction coefficient ($k$)—skyrockets.
As a result, light passing through the device is rapidly attenuated. As Sensong An et al. highlight in Deep neural network enabled active metasurface embedded design, this parasitic absorption creates a fundamental bottleneck, severely restricting the transmission efficiency of tunable optical components.
If a single switch absorbs a significant fraction of the light passing through it, cascading dozens or hundreds of these elements into a complex routing architecture becomes impossible. The signal simply vanishes before reaching its destination, a limitation emphasized by Zhuoran Fang et al. in Non-volatile Integrated Photonics enabled by Broadband Transparent Phase Change Material.
The Solution: Widening the Bandgap
To eliminate this material-induced signal attenuation, researchers have shifted their focus toward a new class of optical PCMs (O-PCMs) engineered specifically for broadband transparency. The core strategy is simple yet powerful: widen the electronic bandgap.
By choosing elements that form wider bandgaps, researchers ensure that the energy of telecommunication (and even visible) photons falls well below the threshold required to trigger band-to-band absorption. Crucially, these materials must still undergo a structural phase transition that alters the real part of their refractive index ($\Delta n$), allowing them to modulate the phase of light without absorbing it.
Several material families have emerged to realize this intuition:
- GSST ($\text{Ge-Sb-Se-Te}$): By substituting selenium for a portion of the tellurium in traditional GST, Yifei Zhang et al. demonstrated in Broadband Transparent Optical Phase Change Materials that the material's bandgap could be opened up, yielding unprecedented broadband transparency and a drastically improved optical figure of merit.
- Antimony Trisulfide ($\text{Sb}_2\text{S}_3$): In a series of foundational implementations, Zhuoran Fang et al. integrated $\text{Sb}_2\text{S}_3$ onto standard silicon and silicon nitride platforms (Low-loss Non-volatile Phase-change Integrated Photonics at 1550nm and 750nm; Non‐Volatile Reconfigurable Integrated Photonics Enabled by Broadband Low‐Loss Phase Change Material). Because $\text{Sb}_2\text{S}_3$ features a wide bandgap in both its amorphous and crystalline states, it provides high-contrast phase modulation with near-zero absorptive losses across both visible and near-infrared spectrums.
- Antimony Selenide ($\text{Sb}_2\text{Se}_3$): Offering a similar wide-bandgap advantage, $\text{Sb}_2\text{Se}_3$ has become a prime candidate for ultra-compact silicon photonic routing due to its exceptionally low insertion loss at telecommunication wavelengths.
Scaling to Complex Photonic Architectures
By mitigating material-induced optical loss, these transparent PCMs change the scalability math for integrated photonics. Systems can now transition from single, isolated components to dense, multi-component networks.
Multi-Bit Storage and Programmable Logic
Because the light is no longer absorbed, researchers can cascade multiple PCM cells together or program a single cell into intermediate, partially crystalline states. For example, Rui Chen et al. demonstrated a 5-bit (32-level) electrically programmable architecture using $\text{Sb}_2\text{S}_3$, maintaining an insertion loss under 1.0 dB while drawing zero static power (Non-volatile electrically programmable integrated photonics with a 5-bit operation; Electrically programmable wide bandgap phase-change material Sb2S3 enabling multi-level non-volatile photonics).
Similarly, Jiawei Meng et al. leveraged these transparent properties to build non-volatile photonic random-access memory (PRAM), preventing the rapid signal degradation that typically occurs when light passes through sequential memory banks (Electrical programmable multilevel nonvolatile photonic random-access memory). On a larger network scale, the authors deployed $\text{Sb}_2\text{Se}_3$ to construct "NEO-PGA," a non-volatile programmable photonic gate array capable of deterministic multi-bit routing without signal degradation (Toward large-scale nonvolatile electrical programmable photonics with deterministic multilevel operation).
Advanced Device Design
Low-loss PCMs also unlock new design paradigms. Hongyi Sun demonstrates that integrating transparent PCMs with Bragg structures allows for non-volatile filter tuning without degrading the resonator's signal integrity (Nonvolatile Tuning of Bragg Structures Using Transparent Phase-Change Materials).
To maximize performance, Maoliang Wei et al. combined algorithmic "inverse design" with $\text{Sb}_2\text{Se}_3$, optimizing ultra-compact mode converters and switches that are highly resilient to fabrication errors (Inverse design of compact nonvolatile reconfigurable silicon photonic devices with phase-change materials).
[ Waveguide Core ] === (Low-Loss PCM Layer) ===> High Quality Factor (Q)
Minimal Insertion Loss
Cascadable Architecture
Furthermore, the utility of these materials extends to rapid prototyping and extreme environments. Forrest R. Miller et al. developed dielectric-assisted PCM waveguides that allow for laser-written, etch-free rewritable circuits with minimal propagation loss (Rewritable Photonic Integrated Circuits Using Dielectric-assisted Phase-change Material Waveguides).
Finally, addressing the strict thermal budgets of quantum computing, Uthkarsh Adya et al. successfully demonstrated non-volatile tuning of silicon micro-ring modulators at cryogenic temperatures below 4 Kelvin (Non-volatile tuning of cryogenic silicon photonic micro-ring modulators). By replacing continuous thermo-optic tuning with wide-bandgap PCMs, they eliminated parasitic heat dissipation while preserving the high quality factors required for quantum interconnects.
By shifting the physical baseline from narrow-bandgap absorption to wide-bandgap transparency, this research direction provides the foundational material framework needed to make zero-static-power, large-scale programmable photonic chips a practical reality.
Key Foundational Work
Non-volatile Integrated Photonics enabled by Broadband Transparent Phase Change Material (2020) — contributed 1 concept elements to this direction.
Broadband Transparent Optical Phase Change Materials (2017) — contributed 1 concept elements to this direction.